電腦滑鼠活動記錄系統開發與測試

Abstract

目的：由於電腦圖形使用介面及網路使用的普及，指示操控已成為電腦作業不可或缺的輸入工具，其中又以滑鼠使用最為普遍。另外，一般肌肉骨骼傷害研究方面所使用的電腦作業評估工具通常較昂貴且體積龐大，不適合應用在職場方面的流行病學調查。有鑑於此，本研究開發滑鼠活動記錄軟體，以期應用於電腦終端機（VDT）作業負荷的評估。 材料與方法：設計Mlog v.1記錄軟體並使用力感測滑鼠記錄以下參數：滑鼠點、放的動作及時間、螢幕上游標移動的相對位置及時間，以及點擊滑鼠的力量。Mlog的使用介面有中英文兩種，可設定鍵盤控制、存檔路徑、存檔週期及取樣頻率等參數。在軟體測試上，以示波器記錄滑鼠所產生的電訊號作為黃金標準，測試滑鼠在1.0 GHz、2.0 GHz及3.2 GHz三部不同電腦，分別開啟0、1、5及10種應用程式的工作負荷狀況下，記錄滑鼠在100 ms、200 ms、300 ms及450 ms（點選）與1000 ms、2000 ms、3000 ms及10000 ms（拖曳）等八種不同壓放時間長度下Mlog v.1記錄的準確度及精確度。另外並測試Mlog v.2軟體在3.2 GHz的電腦及未開啟任何應用程式情形下，記錄滑鼠在100 ms、200 ms、300 ms及450 ms與1000 ms五種不同壓放時間長度下，資料縮減率為1~9倍時，Mlog v.2記錄的準確度及精確度。並利用標準砝碼及ALGOL SV-55S按鍵開關荷重測試儀測試力感應滑鼠的準確度。最後再以力感測器訊號與Mlog v.1的記錄時間作為參考點，測試兩系統在同步監測時記錄時間的誤差。 結果與討論：Mlog v.1記錄值與示波器記錄值之間相差值的平均值與標準差為16.2及22.3 ms，比值的平均及標準差分別為1.03及0.04。記錄值與示波器記錄值相關係數為1.000 ( p<0.001 )。由上述結果可以說明，Mlog v.1可以準確而有效的記錄滑鼠的動作。以線性複迴歸模式進行分析發現，Mlog v.1與示波器的記錄值的差異在三種不同電腦之間是有統計上顯著差異的，但是誤差範圍在2.984 ms內，而在四種不同軟體運作負荷下的記錄值則沒有統計上顯著差異。但是不論是否考慮電腦規格或系統負荷，Mlog的記錄值較示波器記錄值大，平均約有9.41 ms左右的系統性誤差產生。而新版Mlog v.2軟體改善了記錄時間的解析度，經測試後誤差可減少在1 ms以內。力感應滑鼠方面，在0~1000 gf範圍內，力感應器量測值與砝碼值之間相關係數r達0.998，顯示力感測滑鼠可以用來施測滑鼠操作者施於滑鼠的力量大小，並可與表面肌電圖和電子測角器做結合，記錄滑鼠使用者手部動作，探討VDT操作者使用滑鼠的動作與手部肌肉活動的關係。另外，將Mlog與打字活動記錄軟體KBlog結合，可用來作為完整的電腦作業負荷評估工具，驗證肌肉骨骼傷害調查問卷中評估VDT工作者電腦、鍵盤及滑鼠使用時間的信效度。

Objective: Due to the prevailing of internet and graphical user interface, the pointing device has already become an indispensable input device for computer work. Among them, computer mouse is the most common one. Usually, the tools for exposure assessment in musculoskeletal disorder study are very large, and expensive, and not appropriate for field epidemiological study. Thus, we try to develop a mouse activity recording program, Mlog, in order to meet such a need in large-scale VDT work task study. Materials and Methods: The Mlog was programmed to record the following parameters: mouse action, time of the mouse being pressed and released, and the coordinates and their corresponding time of the cursor on the screen. There were both Chinese and English interfaces for Mlog. The Mlog user can set up parameters for keyboard control, filing path, filing frequency, and record sampling rate, etc. In the aspect of program test, we use oscilloscope measurement as a golden standard to compare with the electric signals of computer mouse recorded by Mlog v.1 in order to examined the accuracy and reliability of Mlog under various working conditions, i.e., three types of computer capacity – 1.0 GHz, 2.0 GHz, and 3.2 GHz, respectively, and four kinds of computer work loading – 0, 1, 5, 10 programs being executed, respectively. For each combination of computer working condition, the Mlog v.1 was tested with mouse being pressed for different time periods, i.e., 100 ms, 200 ms, 300ms, 450ms for clicking movement, and 1,000 ms, 2,000 ms, 3,000 ms, 10,000 ms for dragging movement. Mlog v.2 was also developed to improve the accuracy of mouse activities recording. In order to compare the accuracy of Mlog v.2 with that of Mlog v.1, Mlog v.2 was examined for different click or drag time periods, i.e., 100 ms, 200 ms, 300 ms, 450 ms, and 1000 ms on 3.2 GHz computer without program being executed . Mlog v.2 was also examined for different data reduce rate ranging of 1 to 9. In addition, we developed a force sensing mouse to monitor force exerted by mouse user. The accuracy of click time being measured by force sensing mouse was checked with level M5 counterpoise and Silicone Rubber Key Switch Feeling Tester. Results and Discussion: The mean difference between Mlog and oscilloscope measurements was 16.2 (22.3) msec, while the mean (SD) of their ratio was 1.03 (0.04). Meanwhile, the Mlog and oscilloscope measurements was highly correlated with r of 1.000 (p<0.001). Besides, results of multiple linear model analysis showed that computer type has signification effects on the measurement for mouse activity by Mlog v.1, but the difference between computer types was less than 2.984 ms. The new version Mlog v.2 program improved the resolution of the mouse activities recording. The error of Mlog v.2 recording was less than 1 ms. In addition, for the Force sensing resistance) mouse, the force sensor A201 was first tested by level M5 counterpoise with a correlation coefficient, r, of 0.998. FSR mouse was therefore a reliable tool to measure force exerted by operator. It can also work together with Mlog, sEMG and electrogoniometer. Based on the above-mentioned results, it was concluded that Mlog could record the mouse activity accurately, precisely and effectively. The Mlog software developed in the present study could be used in the future study to record the mouse user's movements, and, even more, to probe into the relationship between the computer mouse movements and muscle activity of the user. In addition, it also can be combined with keyboard activities recording program, KBlog, as a more comprehensive exposure assessment tool, to validate the information collected through questionnaire regarding the time of computer, keyboard and mouse use by the VDT users.